Winrates can easily be manipulated on this site already. What's good about this site is that the core is not competititive at all. You can easily have fun on this site without winning a lot. For top players there are good players to play against. That's good too. But you do not have to win to have fun in a game

Edited 4/1/2023 19:56:04

Referring to 2.1, I shall attempt a remedy. I've discussed this before, (your level, rating, skill)

I reckon I have a (possible) solution. First, let's clearly define the three:

Briefly:

-Level - This represents your "seat time"
-Skill - How you perform overall, regardless of template
-Rating - Indication of your abilities given a specific format

Specifically:

Level - Points earned from playing. Leveling up (to 54) can be done in weeks. I and others have. While Achievements and Unlockables are awarded, your level # represents little of your skill or ability. Primarily, it is an indication of seat time or play time if you prefer. Though it is true, you need skill to play and win, thus earning points, however, one need not be elite to level up. My current level of 59 tells you nothing other than I've played often. I'm an average player.

Skill - This is your overall ability to perform successfully in a given situation. Your adaptability, if any, comes into play. Skilled you say? Well, exactly how skilled are you? Sometimes too broad a term.
Example - Martial arts. A brown belt is not as skilled as a black belt, but more skilled than a blue belt. When they face each other in sparring, the outcome is usually a fore gone conclusion. But all three are "skilled". Which leads to...

Rating - This is your ability to perform successfully in specific, repeatable situations. Ratings should be reserved for how well someone performs in a specific, repeatable scenario. If your victories are many, you are said to be, "highly skilled". If your performance is lack luster, the word, highly, is absent, and your just average or less. Your rating in one template may not be the same in another template. But you are/were no more or less skilled when playing on one template versus another.

So I propose this. Let us agree that your Performance is a result of two components: skill and rating. Your rating is how well you do specifically, and your skill is an average of your individual ratings. While this can lead to a wide gap for calculating averages, we simply drop the skill component of a players profile. Judge a players "skill" by the their individual per template "ratings".

This gives a clearer picture of who and what you're dealing with. Your level is cosmetic. Your skill is the contributing component to your rating. Your rating is the relevant element.

To that end, I submit the following. Adopting the method used by MLB,

One of the oldest and most universal tools to measure a hitter's success at the plate, batting average is determined by dividing a player's hits by his total at-bats for a number between zero (shown as . 000) and one (1.000).

Using this ^, I believe we can streamline the system while simultaneously providing increased accuracy.

So, for match making, like Wrestling, match as closely as possible within a given range of weight. In this case, Ratings. No more than a 25 point spread. For example, a player with a Rating of .625 would get matched with someone from .600-.650. That is reasonable.

In case of no human player found, default to AI, so the player is not wasting their time. Nobody likes the sit-spin. Which means we need better AI with the ability to be flexible and adapt accordingly.

In most cases, in reference to newbies vs. OP, the new guy probably gets owned, which sucks, and the OP gets miniscule points, which sucks. In the end, a waste of time for both. Please put a stop to this.

And Warzone is awesome. (But oh my gosh, could be so much better)

@DanWL

if you disagree then there is literally no point trying convince you

- Thankfully, I don't give up on people on people that easy, or else we couldn't agree a lot of the arguments we later managed to agree on (like the past arguments about levels, clan caps etc.). Plus don't worry, I'll never try to put text walls to disturb you or explain my points with insulting, my objective is always to brainstorm for the benefit of community, not to pick petty fights.

Win rate never reveals skill, only rating does

- Firstly, I already made experiments, whether you are aware of it or not, there is a correlation between skill and WR. The causality is low but still high enough to make a significant difference. If you want, we can meet up on discord, I can share my screen while I make 1v1 fights against opponents with <40% WR & >60% WR. We can keep a record of the score and I'm confident that I'll beat players with <40% WR more often as long as sample size is 10.

- Secondly, I already said that the suggestion of separating WR games with rating games is to make WR more reliable, which indicates that I'm aware of WR not very reliable. While you are turning your back on WR because you don't find it reliable enough, I'm trying to fix its reliability.

---

@krinid

You don't control who your opponents are, it's whoever is queued up at the time

- Exactly, that's the main reason why people hates QM ratings. Even if it perfectly represents your rating (let's say its 2000), it will put you against a player with 500 rating since there is no player queued with 2000 rating at that time. I don't know the number of active players in a given time, so I never gone into detail on how to improve matchmaking but if there are enough active players, a condition can be added to QM like a player can only face the queued opponent if the rating difference is less than 500.

But what sample size is enough?

- Depends on how much detail you wanna work with. World of statistics usually starts with the rule of 10 (you need at least 10 sample size with perfect randomness for even the most generalized meaningful result). Most generalized result in a player is just whether they are good/bad. To look for more detail, you usually go with multitudes of 2 as you desire 2 times more detail. So if you wanna look which is better among 2 good players, 20 should be enough. If you wanna look which is better among 2 very good player, 40 should be enough and it goes like that...

---

@graemes

I don’t know why so many on this thread are so eager to reinvent the wheel.

- We're not trying to reinvent the wheel. We are talking about WR's reliability, and whether we should use it as an indication for a player's skill. Reinventing the wheel would be thinking of new ways to determine a player's skill. This talk is more like thinking about our current, already invented wheel's efficiency.

---

@HangFire

For example, a player with a Rating of .625 would get matched with someone from .600-.650. That is reasonable.

Yea, I was thinking the same and basically responded to krinid with the same thing. We can even make this range dynamic. For example let's say your rating is 2000, when you queue for QM, for the first 1 minute, game will only look for queued players with a rating with <100 difference. After 1 minute, if you didn't match with anyone, it'll look for <300 rating difference, if still not matched with anyone, it'll look for <900 rating difference...

Which means we need better AI with the ability to be flexible and adapt accordingly.

Also agree. Even tough using AI as opponent is not desirable, if they are improved enough, AIs have the potential do decrease game's dependency on active players. Current AI is already good enough for new players (so you can put them against AI in QM if another new player is not found) but the current problem is about good players. An AI that can satisfy good players won't be easy to make but I believe the time & effort that's gonna be require on improving AIs will be worth the investment

As I said above, win rate does not indicate skill. If you wanted win rate to be meaningful and not a fun statistic, how would you go about it? Rating is to indicate skill, not win rate. And the more games played, the more stable the rating is. There is no direct correlation between win rate and skill. Just like how there’s no relation between level and skill.

The way to improve the AI is to let it know about the games history (this isn’t accessible as far as I know). Upgrading the AIs will break the existing AIs so a new AI will need to be released (and not replace any existing ones)

@VERAVARI

.....my objective is always to brainstorm for the benefit of community,...

Respect. +1

+1 graemes

I think it would contribute to Warzone's popularity if TLA changed their logo back

Agreed, the TLA logo change has very negatively affected the ratings of warzone.

I've seen a one-star Google review from a person who said they played the game for nine years and now quit because they couldn't handle the lack of any aestehics of the new TLA logo and now their life makes no more sense

tbh I consider quitting every time I look at the TLA logo too

And recently I saw this for the first time and was devastated to find out that our lovely Python clan jacked their logo from the Python (programming language)

why are you posting swastikas again, krinid?

you mean Python is a nazi clan? who knew?

I like how the original version has Ukraine colours

So what's our takeaway from this

my takeaway is that python will vote for kanye is 2024

my takeaway is

'Warzone'.split('').join('     ');

Edited 4/6/2023 22:22:34

For one, better tutorials for new players would be nice. Secondly, having a larger team working to improve the game would be very beneficial. Make the website look more official than it does now, and making it more intuitive, should go a long way I believe. Adding new features isn't attracting new players; but the lack of intuitiveness for new players turns the newbies away. I hope to see this game thrive like this in the future.

Not to say we don't like new features, WE LOVE NEW FEATURES

Lmao krinid I forget myself how much people don't know about programming languages

...I forget myself how much people don't know about programming languages

You could teach us. You have your fan base and some of us would love to learn from you. ;)